fd-portfolio-rollup
Skill1 starsChanged 4 months agoMITView on GitHub
The one-screen Thursday portfolio picture across every Forward Deploy firm Jeff covers (see `_shared/fd-firms.md` for the current firm list). Reads each firm's engagement-state memory file plus its latest delivery trajectory, flags stale state and gate-blocked firms, separates program-level escalations from pod work, and recommends where the week's attention goes. Sol dispatches Vantage; Sol reviews and delivers to the shared Program/00_delivery/status/ folder. Trigger on "portfolio roll-up," "Thursday roll-up," "where are my firms," "portfolio status," "cross-firm status." It recommends a ranking; it never decides where the week actually goes, and it never presents stale state as current.
In this repository13 other files it gives its agents.
- CLAUDE.mdCLAUDE.md
- fd-blocker-sweepSkill
- fd-firm-onboardSkill
- fd-onsite-availabilitySkill
- fd-reconcileSkill
- full-sweepSkill
- intakeSkill
- process-meetingsSkill
- safe-mvSkill
- slack-recapSkill
- steSkill
- sweep-binderSkill
- sweep-jiraSkill
The file
We split this skill into the rules it gives an agent
7 rules. About 0 are common, 6 are distinctive, and 1 is very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈6
Distinctive · estimate
This author's own choices. Worth a look.
1
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 26
Default cadence is Thursday morning, before the cross-firm sync.
- Distinctive: line 81
Default scope is every firm found in Step 0.
- Distinctive: line 163
No firm silently missing: every firm from Step 0 is either a real row or a gap row.
- Distinctive: line 165
No decision made or reworded.
- Distinctive: line 166
No firm bleed: each row is built only from that firm's own files (information-wall rule).
- Distinctive: line 168
No padding: quiet firms are allowed to say "nothing pressing."
- Very specific: line 189
Put <!-- audience: internal --> as the first line — this is an internal portfolio picture, not external firm-facing prose, so it's exempt …
≈ “Common” is an estimate: at least 1 in 5 of the 372,429 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
<!-- audience: internal -->FD Portfolio Roll-Up
The cross-firm view a DM takes into the Thursday sync: where every firm sits on the delivery spine, which are gate-blocked, what needs escalation above the pod, and where attention goes this week. Sol orchestrates and reviews; Sol never does the aggregation itself. Sol dispatches Vantage (the Forward-Deploy Firm Intelligence Lead) to do the reasoning, then reviews and delivers. It never decides the week for Jeff.
Sourced from FD-P-13 (Playbook Ch 8). Ported into Sol from the retired standalone
FD starter-kit. The biggest path change from the starter-kit: per-firm state is no
longer a firm-local firm_state.md; it lives in Sol's assistant memory, one
file per firm (see Step 0).
When to run
- Jeff asks for a "portfolio roll-up," "Thursday roll-up," "where are my firms," "portfolio status," or "cross-firm status."
- Default cadence is Thursday morning, before the cross-firm sync.
- Scope override Jeff can state: a subset of firms ("just Aprio and CLA"). Default is every FD firm with an engagement memory file.
Step 0 — Locate per-firm state (Sol's memory) and pre-check staleness
This skill DEGRADES, it does not refuse. Run it even when state is stale or a firm file is missing; report the gaps honestly instead of stopping.
Per-firm engagement state lives in Sol's assistant memory, one file per firm:
/Users/jeffheinen/.claude/projects/-Users-jeffheinen-sol/memory/<firm>-forward-deploy-engagement.md(e.g. aprio-forward-deploy-engagement.md is the live example — read it for the
shape.) Jeff's current FD firms are listed in
_shared/fd-firms.md — read it for the current list; a firm
without an engagement memory file yet is a gap row, not an invented one.
- Glob that memory dir for
*-forward-deploy-engagement.md. Each file is one firm in the portfolio. - Staleness check (in-skill, N = 7 days). For each engagement file, establish
its freshness date in this order: (a) an explicit
as of: YYYY-MM-DDline in the body, if present; else (b) the most recent date the file's content asserts about the engagement; else (c) the file's own last-modified date as a last resort, noted as such. Prefer a content date over file mtime (a drive sync can touch mtime without any content change). If the established date is more than 7 days before today, or no date can be established, mark that firm STALE. 7 is the natural unit: the cadence is weekly and FD-P-13 flags any firm state more than a week stale. - A firm named in scope but with no engagement memory file (or with the file but no delivery trajectory to read) becomes a gap row, never a silently dropped firm.
- State the staleness and gap findings in a one-line pre-dispatch summary before you dispatch (e.g., "Aprio current; CLA STALE (as of 2026-07-02); Sorren no engagement memory file yet — gap row").
Delivery trajectory. The most recent trajectory line per firm comes from that firm's latest delivery status in its Intuit-internal Drive firm repo — the canonical FD Shared Drive firm path (quote it, trailing space after "Forward Deploy"):
/Users/jeffheinen/Library/CloudStorage/GoogleDrive-jeff_heinen@intuit.com/Shared drives/MM Partner Program Central Drive/Forward Deploy /<Firm>/Firms sit flat directly under that path — NO extra Forward Deploy/ wrapper inside
<Firm>/. Check 00_delivery/status/, 00_delivery/blockers/, or
02_discovery/readouts/ — if the folder is mounted. If a firm's folder is not
mounted, do NOT fail the whole roll-up — note "trajectory unavailable (firm folder
not mounted)" for that firm and proceed on the memory file alone. (Unlike
fd-reconcile and fd-blocker-sweep, the roll-up degrades on a missing mount rather
than stopping, because a Thursday picture with caveats beats no picture.) See memory
drive-folder-routing for the canonical layout.
Step 1 — Resolve scope
Default scope is every firm found in Step 0. If Jeff named a subset, use it. State the resolved scope in one line and proceed; do not ask.
Step 2 — Dispatch Vantage (one subagent for the whole roll-up)
Sol spawns a single Vantage subagent for the whole roll-up (not one per firm): the inputs are short curated state files and cross-firm ranking is the point, so a single agent that sees all firms at once is correct.
Model tier: claude-sonnet-5 (mid-tier). This is structured aggregation plus
ranking over short curated files, not messy-corpus synthesis. Escalate to
claude-opus-4-8 only if 2 or more firms show conflicting or ambiguous state that
needs real adjudication to even describe.
Paste each firm's engagement-memory content and its latest trajectory line inline into the spawn prompt (they are short; do not make Vantage hunt for them). The spawn prompt must be fully self-contained. Include ALL of the following.
Who Jeff is (read _shared/fd-firms.md and paste its
"Who Jeff is" block verbatim into the spawn prompt — prevents the recurring
role/firm errors):
Identity and grounding discipline (paste verbatim):
You are Vantage, doing a meticulous Forward Deploy portfolio roll-up. You never fabricate. Every factual claim traces to a firm's engagement memory file or its delivery trajectory, cited with the file and its state date. Anything you infer is tagged [Unverified] inline. A short, honest row is correct: "nothing pressing" is a valid cell, and you do not pad a quiet firm. You end the deliverable with the trace footer specified below.
House rules (paste verbatim):
- Grounding: every factual claim cites source file plus date. Untraceable claims are tagged [Unverified] inline. End with the trace footer.
- Roles, not names, in anything that could travel to a firm.
- Surface, never decide: you propose a ranking; you do not decide where the week goes, and you do not resolve or reword any open decision.
- Regulated-data restraint: any regulated or sandbox-data item is flagged "escalate, do not action," never summarized or reasoned to a conclusion.
Per-firm inputs (paste inline): for each firm, its full engagement-memory content, its most recent delivery trajectory line (or "trajectory unavailable — firm folder not mounted"), and its STALE or gap flag from Step 0.
Output template Vantage fills:
One table row per firm, columns: Firm / pod | Spine phase (Week 0 / Discovery / Design / Prototype / Build & Finetune / Deploy / Enablement — per-agent status from Phase 3 on) | Trajectory (on track / at risk / slipping) | Single most pressing open item + owner role + agent threatened | State date (append STALE if flagged). A firm with nothing pressing says so in the cell. A gap row names the missing file rather than inventing state.
Then three sections:
- Gate-blocked firms — each firm in Design with an unresolved conflict preventing scope lock, with the blocking decision named.
- Program-level escalations — items above the pod (regulated-data or sandbox question, slipping SOW, missing platform dependency). These need escalation, not pod effort. Do not age them; list them.
- Attention call for the week — the firms ranked by urgency (not by phase), each with a one-line reason. This is a recommendation for Jeff to confirm, never a decision.
End with the trace footer:
--- **Sources:** [each engagement memory file and its state date; each trajectory line] **Claims marked [Unverified]:** [list, or "none"]
Judgment boundaries (paste verbatim):
You must NOT: decide where Jeff's week actually goes (you recommend a ranking, Jeff decides); present stale state as current (carry the STALE flag into the row); mix one firm's specifics into another firm's row beyond the shared table columns; resolve or reword any open decision.
Output contract for Vantage: return the full roll-up content to Sol as its result. Vantage does NOT write the file — Sol delivers it (Step 4).
Step 3 — Sol review
Before delivering, check:
- Trace footer present; every row's state date cited; STALE flags carried through from Step 0 (Vantage must not have quietly dropped one).
- No firm silently missing: every firm from Step 0 is either a real row or a gap row.
- No decision made or reworded; the attention call reads as a recommendation.
- No firm bleed: each row is built only from that firm's own files (information-wall rule).
- No padding: quiet firms are allowed to say "nothing pressing."
Return to Vantage for revision if any check fails.
Step 4 — Deliver (Sol writes to the shared Program status folder)
The roll-up is a program-level artifact (cross-firm), so per the personal-vs-shared
rubric it goes to the shared repo, NOT owners_inbox/. Sol writes it to the
canonical FD Shared Drive (quote it, trailing space after "Forward Deploy"):
/Users/jeffheinen/Library/CloudStorage/GoogleDrive-jeff_heinen@intuit.com/Shared drives/MM Partner Program Central Drive/Forward Deploy /Program/00_delivery/status/fd-portfolio-rollup_<YYYY-MM-DD>.mdTest writability first (ls -ld the status/ folder; drwx = OK) and fail loudly
if the Program folder is absent or read-only. Use the Write tool. (Old retired
targets: owners_inbox/, a firm Drive folder, a _portfolio/rollups/ folder, and the
fully-retired My Drive/Agent Studio Forward Deploy/ tree — superseded 2026-07-24.)
See memory drive-folder-routing for the rubric and current layout.
Hook note: because it no longer lands in owners_inbox/, the check-trace-footer.py
and check-humanizer.py hooks do NOT cover it — Sol's Step-4 review enforces the
trace footer instead (required by house rules regardless).
- Put
<!-- audience: internal -->as the first line — this is an internal portfolio picture, not external firm-facing prose, so it's exempt from the humanizer pass (per CLAUDE.md's internal-artifact rule). Only humanize if Jeff says he'll forward it.
Give Jeff a tight chat summary: the ranked attention call, the gate-blocked firms, and the program-level escalations. Name any STALE or gap firms explicitly so Jeff knows what to refresh.
What this skill must NOT do
- Have Sol do the aggregation. Sol orchestrates; Vantage aggregates.
- Decide where Jeff's week goes. It recommends a ranking; Jeff owns the call, confirming it against what no engagement memory file captures.
- Present stale state as current. Stale firms are flagged STALE, never silently passed through as fresh.
- Mix one firm's specifics into another firm's row beyond the shared table.
- Resolve, close, or reword any open decision on any firm.
- Refuse to run because state is stale, a firm file is missing, or a firm folder is not mounted. It degrades honestly (STALE flags, gap rows, "trajectory unavailable"), because a Thursday picture with caveats beats no picture.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

